[P2][Logs] Reduce empty-cluster lifecycle noise during startup and shutdown

Parent issue: #4691

Intent

In an empty cluster, startup and shutdown currently produce many lifecycle logs that mostly describe module sequencing. They become especially noisy during frequent restarts or control-plane flaps. This task keeps the key checkpoints but collapses routine lifecycle chatter into concise startup and shutdown summaries.

Modification points

• Review server, election, HTTP, and gRPC module lifecycle logs for empty-cluster startup and shutdown.
• Replace scattered starting, closing, and closed narration with a small number of summary logs for startup and shutdown phases.
• Keep detailed logs only for failures, unusual delays, or diagnostics mode.

Entry points

• server/server.go:296
• server/server.go:303
• server/server.go:315
• server/server.go:332
• server/server.go:422
• server/server.go:450
• server/module_election.go:90
• server/module_http.go:75
• server/module_grpc.go:67

Done criteria

• Empty-cluster startup and shutdown emit materially fewer routine lifecycle logs.
• The key control-plane transitions are still visible from the default log path.
• When a failure occurs, the logs still make the failing phase obvious.

Start by reviewing the lifecycle logging at the listed server/server.go locations and in server/module_election.go, server/module_http.go, and server/module_grpc.go. Reproduce or inspect empty-cluster startup and shutdown logs first, then verify that routine narration is condensed while key transitions and failure phases remain visible by default.